Image may be NSFW.
Clik here to view.
Think of a situation where you want to ‘see more’ than your eyes can reveal; to spot the early signs of mildew or aphid damage in crops, or to discover the health and condition of your skin. The human eye has three types of colour receptors, cones, centred roughly around 430, 540 and 570 nm wavelengths, our brain combines these into what we perceive as colour. But this isn’t the only way to see colour: most mammals have just two types of cones, whilst bees are able to see right into the UV, allowing them to home in on the centre of flowers but costing them the ability to see anything past yellow. We’re all used to colour cameras, which split light into red, green and blue to simulate what your eyes see, but now technology gives us the chance to extend our colour vision beyond our human biology and see the world with any combination of wavelengths we choose. The technique I’m describing is multispectral imaging, or taken to its extreme, hyperspectral imaging.
Multispectral imaging is the capture of images in which each pixel has a value for a number of wavelengths – this could range from just a few broad wavelength ranges right up to capturing a full spectrum of narrow bands, at which point the technique is called hyperspectral imaging. Optical filters can be used to see each waveband in turn, or by controlling the illumination and pairing this with an algorithm to extract the useful data, a normal camera can see each of the desired wavebands simultaneously. The wavebands can be chosen at particular points to reveal otherwise hidden information about the object being imaged. You can even calculate the relative absorption at different wavelengths to perform imaging spectroscopy, and find out about the composition of the material being imaged. For example, you could choose wavelengths to separate the pattern of blood just below the skin surface from the melanin and keratin that would otherwise obscure it.
The beauty of this technique is its simplicity. A cheap silicon camera of the sort you’d find in any smart phone has all the technical capability you need to reveal the hidden world of your choice, from UV through the visible spectrum, and into the near infra-red. Add to this some carefully chosen but dirt cheap LEDs, and an algorithm to extract the signal at each wavelength, and you’ve made a simple imaging spectrometer tailored precisely to your needs. If you compare this to a full-scale hyperspectral camera, you’ve avoided collecting vast quantities of useless data and paying thousands of pounds for the privilege. The same technique could also push beyond the limits of a standard silicon camera, and use other semiconductor detectors to probe deep into the infrared, giving insights into a world of ‘colour’ hidden right before our eyes.
